FOREX trading strategy optimization

Abstract

Developing robust trading rules for forex trading remains a significant challenge for both academics and practitioners. We employ a genetic algorithm to evolve a diverse set of profitable trading rules based on weighted moving average method. We use the daily closing rates between four pairs of currencies – EUR/USD, GBP/USD, USD/JPY, USD/CHF – to develop and evaluate our method. Results are presented for all four currency pairs over the 16 years from 2000 to 2015. Developed approach yields acceptably high returns on out-of-sample data. The rules obtained using our genetic algorithm result in significantly higher returns than those produced by rules identified through exhaustive search.